Barcelona Opteron 2350(B1) arrived

Printable View

Show 100 post(s) from this thread on one page

11-09-2007, 07:47 AM
nemrod

Quote:

Originally Posted by Sunfire

Maybe I just missed it, but how is the memory clock calculated?

Is it what you ask:

Quote:

Originally Posted by cpuz

Hi Kyosen,
I got your PM, we'll work on that.
Everest is assuming that memory clock is obtained from bus clock on K10, and CPU-Z computes it from CPU clock (as for K8).
IMO Everest is correct, but Tamas & I had no confirmation at the moment. We're expecting a benchmark can tell what the real clock is.

Nice rig, I'm impatient to see results on the new BA stepping :up:
11-09-2007, 08:14 AM
Sunfire

Ok, memory clock is obtained from bus clock but how?
So how you get 333MHz for DDR2-667 on Barcelona, and 400MHz for DDR2-800 or 533MHz for DDR2-1066 for Phenom?
11-09-2007, 08:31 AM
tictac

Quote:

Originally Posted by Sunfire

Ok, memory clock is obtained from bus clock but how?
So how you get 333MHz for DDR2-667 on Barcelona, and 400MHz for DDR2-800 or 533MHz for DDR2-1066 for Phenom?

i came out with this.... but no confirmation yet :shrug:

Memory Clock Speed

Code:

NB Speed HTT DDR400 DDR533 DDR667 DDR800 DDR1066 1600Mhz 200MHz 200MHz 266MHz 320MHz 400MHz 533MHz 1800MHz 200MHz 200MHz 257MHz 300MHz 360MHz 450MHz

Memory clock = NB Speed / memory divider

Memory divider
DDR 400 = NB Multiplier
DDR 466 = NB Multiplier - 1
DDR 533 = NB Multiplier - 2
DDR 667 = NB Multiplier - 3
DDR 1066 = NB Multiplier - 4

indiana_74... few more tweak to set your HT link speed and NB Speed
Link : http://www.xtremesystems.org/forums/...=164768&page=2
11-09-2007, 10:35 AM
Sunfire

Quote:

Originally Posted by tictac

i came out with this.... but no confirmation yet :shrug:

Memory Clock Speed

Code:

NB Speed HTT DDR400 DDR533 DDR667 DDR800 DDR1066 1600Mhz 200MHz 200MHz 266MHz 320MHz 400MHz 533MHz 1800MHz 200MHz 200MHz 257MHz 300MHz 360MHz 450MHz

Memory clock = NB Speed / memory divider

Memory divider
DDR 400 = NB Multiplier
DDR 466 = NB Multiplier - 1
DDR 533 = NB Multiplier - 2
DDR 667 = NB Multiplier - 3
DDR 1066 = NB Multiplier - 4

indiana_74... few more tweak to set your HT link speed and NB Speed
Link : http://www.xtremesystems.org/forums/...=164768&page=2

This is interesting. Thanks for the info!
The cores and the NB are separated from each other, so the voltages can be adjusted separated too. With a locked Phenom where you can't raise the multiplier just the HT freq, the NB freq is going with it? Or it can be controlled separetly too? And the L3 and Ram freq?
11-09-2007, 10:55 AM
dess

Quote:

Originally Posted by Sunfire

With a locked Phenom where you can't raise the multiplier just the HT freq, the NB freq is going with it? Or it can be controlled separetly too? And the L3 and Ram freq?

It would go with it, but probably you can degrade the NB's own multiplier. (Hopefully AMD won't lock it downwards. edit: would be a stupid idea.)
L3 freq is the same as NB's, isn't it?

BTW, HT freq is already a derived clock, using the HT multiplier - you mean the external clock/system clock. (HT clock = ext. clock * HT multiplier.)
11-09-2007, 11:14 AM
informal

Quote:

Originally Posted by dess

It would go with it, but probably you can degrade the NB's own multiplier. (Hopefully AMD won't lock it downwards. edit: would be a stupid idea.)
L3 freq is the same as NB's, isn't it?

BTW, HT freq is already a devided clock, using the HT multiplier - you mean the external clock/system clock. (HT clock = ext. clock * HT multiplier.)

"L3 freq is the same as NB's, isn't it?"-> yes and OCing the whole NB would easily bring down latency of L3 and mem. latency.Now the NB runs at rather low 1.8Ghz.WIth the NB clock up to 2.5Ghz,we could see a nice boost in some cache sensitive apps.Also,every single Phenom we saw had its ram clocked low with poor timing :(.I can't believe people who could get their hands on such a sparse part at this moment ,couldn't get some decent LL DDR2 kit :shrug:
11-10-2007, 04:03 AM
cpuz

Quote:

Originally Posted by tictac

Memory clock = NB Speed / memory divider

Hi Tictac !

Are you use the memory clock is computed from NB speed ?
I'd rather assume that :
Memory clock = HTT Speed / memory divider
But it's just a guess ...

If you're correct, that would mean that memory clock also depends on the single/dual plane setting (the NB speed depends on it).
11-10-2007, 04:14 AM
tictac

yeah.. memory speed computed from nb speed.. ht link speed use from nb to chipset or other cpu.. nb speed use to bridge nb to the 4cpu core... so i go with nb speed..
11-10-2007, 06:35 AM
QuietIce

Quote:

Originally Posted by justapost

So SRI, XBAR, MCT, L3 DCT0 and DCT1 all run on NB speed?
Or do the DCT's run on ram clock?

If I'm understanding the AMD K10 Dev Guide correctly the DCTs run on the NB at NB speed.

As a side note for video BMs: Setting the DCT/DRAM to 128-bit mode will not support 32-byte bursts - only the 64-bit mode will support 32-byte bursts. May make a big difference when running video benches ...

.
11-10-2007, 08:42 AM
Quintero

Quote:

Originally Posted by QuietIce

As a side note for video BMs: Setting the DCT/DRAM to 128-bit mode will not support 32-bit bursts - only the 64-bit mode will support 32-bit bursts. May make a big difference when running video benches ...

Yes, K10 won't support 32-byte granularity in dual channel (single controller), just like any other AMD DDR2 rig. Of all the K8 systems only socket 939 rev E supports 32-byte mode, but none of the earlier 939 revisions or any AM2 CPUs do. Remember that the channels are accessed in perfect parallel (lockstep), and that DDR2 has a minimum prefetch width of 4 columns.

1 4n-prefetch x 2 ch = 8 columns = 64 bytes.
11-10-2007, 10:55 AM
QuietIce

On the K10s unganged mode (2x 64-bit independent) supports the 32-byte burst, ganged mode (2x 64-bit parallel) does not. That might explain some discrepancies in the benches.

Bits and bytes - original post edited ... :)
11-10-2007, 10:02 PM
kyosen

Final update from my KFSN4-DRE

I tried Vcore&Vnb mods, and got 2.6G@Vcore=1.38V, Vnb=1.35V.
I pushed it further, and saw 2.67G SuperPI1M run and then,
the system freezed...it's dead as prepared for arrival of Intel 45nm:hitself:

Last screenshot:
http://www.oohashi.jp/c-board/file/S...K10B1-2.6G.png
11-10-2007, 10:26 PM
Charles Wirth

Sad news indeed.

One or both cpu's dead? mobo?
11-10-2007, 11:19 PM
kyosen

Quote:

Originally Posted by FUGGER

Sad news indeed.

One or both cpu's dead? mobo?

CPU is safe, still alive:)
Just mobo is dead...
...one night removal of CMOS backup battery can't help her:shrug:
11-10-2007, 11:44 PM
STEvil

digital vrm per chance?
11-11-2007, 08:03 AM
informal

Quote:

Originally Posted by kyosen

CPU is safe, still alive:)
Just mobo is dead...
...one night removal of CMOS backup battery can't help her:shrug:

Damn,bad luck man,sorry for your mobo.How is Asus quadFX mobo doing?Is "she" alright :D?
11-11-2007, 11:54 AM
cdawall

when you get your rig going again any chance we could get a sciencemark run?
11-13-2007, 01:47 PM
Boschwanza

@kyosen

Due to the optimization of Cinebench 10 for the Intel Core architecture, is it possible to run a test on Cinebench 9.5 to get a fair comparision between K10 and Core2Q ?
11-13-2007, 02:01 PM
JohannesRS

Quote:

Originally Posted by Boschwanza

@kyosen

Due to the optimization of Cinebench 10 for the Intel Core architecture, is it possible to run a test on Cinebench 9.5 to get a fair comparision between K10 and Core2Q ?

Wait! That's something I wasn not aware of. What heck of optimization for Core2 does Cinebench 10 implements??
11-13-2007, 02:12 PM
Boschwanza

-------------------------------------------
"One major improvement in CINEMA 4D has been the fine-tuned optimization for the Intel® Core™ microarchitecture, which took advantage of MAXON’s ten-years experience in multi-threading. [...] On the next-gen Intel® Core™ 2 processor codenamed Yorkfield, CINEMA 4D performs twice as fast as on an Intel® Core™ 2 Duo processor. CINEMA 4D takes advantage of various multi-core processors from Intel, e.g. Intel® Xeon®, Intel® Core™ 2 Duo and Intel® Core™ 2 Quad processors."

"In order to fully exploit the power of Intel processors, MAXON keeps improving its software code using Intel® C++ Compiler. [...] The Intel®
VTune™ Performance Analyzer and Intel® Threading Tools support MAXON’s developers in controlling performance improvements and identify performance
related issues.

Thanks to the accurate software tuning and the latest Intel technology advancements, CINEMA 4D now runs more than twice as fast as it did only one year ago - comparing the CINEMA 4D performance on the next-gen Quad-Core processor codenamed "Yorkfield" and on the Intel® Core™ 2 Extreme X6800 Dual-Core processor running at the same frequency - as measured by MAXON's benchmarking tool CINEBENCH."

-------------------------------------
11-13-2007, 03:13 PM
haylui

My finding on benchmarking is that certainly the software optimization is a little bias. Intel always releasing new C++ compiler which can greatly optimized for their own processor. Performance gain is doubtful (real performance gain vs scores gain).

Anyway, it's like back into the old day of AMD which a few benchmark can keep up with Intel but the price are much low. I don't see running an AMD platform is like riding a car traveling at 60km/h while Intel's platform is at 60mph.
11-13-2007, 03:26 PM
QuietIce

Quote:

Originally Posted by Boschwanza

-------------------------------------------
"One major improvement in CINEMA 4D has been the fine-tuned optimization for the Intel® Core™ microarchitecture, which took advantage of MAXON’s ten-years experience in multi-threading. [...] On the next-gen Intel® Core™ 2 processor codenamed Yorkfield, CINEMA 4D performs twice as fast as on an Intel® Core™ 2 Duo processor. CINEMA 4D takes advantage of various multi-core processors from Intel, e.g. Intel® Xeon®, Intel® Core™ 2 Duo and Intel® Core™ 2 Quad processors."

"In order to fully exploit the power of Intel processors, MAXON keeps improving its software code using Intel® C++ Compiler. [...] The Intel®
VTune™ Performance Analyzer and Intel® Threading Tools support MAXON’s developers in controlling performance improvements and identify performance
related issues.

Thanks to the accurate software tuning and the latest Intel technology advancements, CINEMA 4D now runs more than twice as fast as it did only one year ago - comparing the CINEMA 4D performance on the next-gen Quad-Core processor codenamed "Yorkfield" and on the Intel® Core™ 2 Extreme X6800 Dual-Core processor running at the same frequency - as measured by MAXON's benchmarking tool CINEBENCH."

-------------------------------------

That certainly holds true for this hardware comparison of Barcelona 1.9GHz v Xeon 1.86GHz:
http://www.hardwarezone.com/articles...?cid=2&id=2411
11-13-2007, 03:29 PM
nemrod

Quote:

Originally Posted by Boschwanza

@kyosen

Due to the optimization of Cinebench 10 for the Intel Core architecture, is it possible to run a test on Cinebench 9.5 to get a fair comparision between K10 and Core2Q ?

Where did you see a fair comparision. Not using fully capabilities of new architecture is suppose to show something or you believe maxon have just discovered vtune?
look at slide around 25:
http://www.securitytechnet.com/resou...2/PCS027PS.pdf

So the conclusion is if you're interessed in CINEMA 4D performance you take the fastest and compare plateforms. If not, and your case is to do something really fair, you do your own code optimized on one side for intel next gen cpu and one the other side for amd next gen cpu. :shrug:
11-13-2007, 03:46 PM
nemrod

Quote:

Originally Posted by haylui

My finding on benchmarking is that certainly the software optimization is a little bias. Intel always releasing new C++ compiler which can greatly optimized for their own processor. Performance gain is doubtful (real performance gain vs scores gain).

Is it doubtful?
http://www.trustedreviews.com/images...e/5887-mp3.gif
11-13-2007, 05:01 PM
STEvil

Used to be a program that would disable Intel processor checks to allow optimized code to run on processors other than "GenuineIntel" ... might help, might not..

EDIT

Wow, the harper and clover system in that hardwarezone writeup are like apples and oranges... 8x 533mhz FB-DIMM's in the Clover and 4x 667mhz FB-DIMM's in the Harper.. best to use 4 sticks (8 is slower) and not matched for speed at all..

Show 100 post(s) from this thread on one page